KanjuTech's Transcription and Diarization model ensures secure end-to-end recognition of multi-participant conversations. It converts dialogue records into precise transcripts with labeled speakers and lines, offering automatic detection for any number of participants. With low error rates (WER and DER) for real-life data, it supports ten languages with human-level accuracy (WER 3-8%). The model efficiently handles over 12 hours of recording in just one hour on ml.p3.2xlarge. Following AWS's secure policy, only users can access processed data through SageMaker products. Industries like translation, transcription, media, broadcasting, call centers, corporate governance, and education will find our solution invaluable for enhancing their products and services.